HybridDAT: A Hybrid Data Association Framework for Robust Online Multi-Object Tracking


Video not available.

Rendering of new sequences is currently deactivated due to heavy load.

Rendering of new sequences is currently deactivated due to heavy load.

Rendering of new sequences is currently deactivated due to heavy load.

Rendering of new sequences is currently deactivated due to heavy load.

Rendering of new sequences is currently deactivated due to heavy load.

Rendering of new sequences is currently deactivated due to heavy load.

Rendering of new sequences is currently deactivated due to heavy load.

Rendering of new sequences is currently deactivated due to heavy load.

Rendering of new sequences is currently deactivated due to heavy load.

Rendering of new sequences is currently deactivated due to heavy load.

Rendering of new sequences is currently deactivated due to heavy load.

Benchmark:

MOT15 |

Short name:

HybridDAT

Detector:

Public

Description:

Global optimization algorithms have shown impressive performance in data-association based multi-object tracking, but handling online data remains a difficult hurdle to overcome. In this paper, we present a hybrid data association framework with a min-cost multi-commodity network flow for robust online
multi-object tracking.We build local target-specific models interleaved with global optimization of the optimal data association over multiple video frames. More specifically, in the mincost multi-commodity network flow, the target-specific similarities are online learned to enforce the local consistency for reducing the complexity of the global data association. Meanwhile, the
global data association taking multiple video frames into account alleviates irrecoverable errors caused by the local data association between adjacent frames. To ensure the efficiency of online tracking, we give an efficient near-optimal solution to the proposed min-cost multi-commodity flow problem, and provide the empirical proof of its sub-optimality. The comprehensive experiments on real data demonstrate the superior tracking performance of our approach in various challenging situations.

Reference:

M. Yang, Y. Jia. A Hybrid Data Association Framework for Robust Online Multi-Object Tracking. In IEEE Transactions on Image Processing, 2016.

Last submitted:

May 30, 2016 (8 years ago)

Published:

August 24, 2017 at 14:20:15 CET

Submissions:

2

Project page / code:

n/a

Open source:

No

Hardware:

3.4 GHZ, 1 Core

Runtime:

4.6 Hz

Benchmark performance:

Sequence MOTA IDF1 HOTA MT ML FP FN Rcll Prcn AssA DetA AssRe AssPr DetRe DetPr LocA FAF ID Sw. Frag
MOT1535.047.735.182 (11.4)304 (42.2)8,45531,14049.378.237.133.341.069.238.160.476.31.5358 (7.3)1,267 (25.7)

Detailed performance:

Sequence MOTA IDF1 HOTA MT ML FP FN Rcll Prcn AssA DetA AssRe AssPr DetRe DetPr LocA FAF ID Sw. Frag
ADL-Rundle-130.648.837.0922,8273,58261.566.937.736.640.669.747.852.076.15.752162
ADL-Rundle-334.143.834.2681,5605,08949.976.533.934.736.973.540.662.280.62.54982
AVG-TownCentre29.244.431.521985324,46537.583.537.726.644.061.428.363.073.51.261246
ETH-Crossing24.141.427.42162973127.190.436.220.838.076.221.571.778.00.1113
ETH-Jelmoli41.856.940.18152731,19353.083.143.337.351.163.641.465.076.70.61049
ETH-Linthescher27.137.728.610139876,40928.296.737.721.839.777.022.376.580.00.11252
KITTI-1646.365.041.21319870658.583.444.138.548.466.243.762.375.70.91033
KITTI-1927.553.436.77131,3102,53552.668.241.433.145.564.339.851.671.71.227184
PETS09-S2L247.746.032.2548174,11257.387.126.339.629.660.742.764.973.71.9110349
TUD-Crossing73.370.451.0824224477.995.347.754.761.259.758.771.977.70.2834
Venice-137.153.639.7547802,07454.576.144.435.647.275.842.459.278.41.71863

Raw data: